评测大模型对物体运动因果的空间推理能力，发现其远低于人类表现。
CausalSpatial: A Benchmark for Object-Centric Causal Spatial Reasoning
- 构建四类任务的诊断基准，测试模型预测物体运动后果的能力。
- 人类准确率84%，GPT-5仅54%，暴露出模型严重依赖语言推理偏差。
- 提出COW框架，通过生成动态视频让模型基于视觉证据进行因果推理。
人类能从静态场景中即时预测后续事件——移动某个物体是否会引发碰撞？这种能力称为因果空间推理。然而当前多模态大模型（MLLMs）仍局限于静态空间感知，难以回答3D场景中的“如果…会怎样”问题。我们提出CausalSpatial，一个诊断性基准，用于评估模型在碰撞、兼容性、遮挡和轨迹四类任务中预测物体运动后果的能力。结果揭示显著差距：人类得分84%，而GPT-5仅54%。分析表明，模型过度依赖文本链式思考，脱离视觉证据，产生流畅但空间不成立的幻觉。为此，我们提出因果物体世界模型（COW），通过生成假设动态的视频，将模拟过程外部化。借助显式的视觉因果线索，COW使模型推理更贴近物理现实，而非语言先验。数据集与代码已公开于https://github.com/CausalSpatial/CausalSpatial。
原文摘要 · Abstract (English)
Humans can look at a static scene and instantly predict what happens next -- will moving this object cause a collision? We call this ability Causal Spatial Reasoning. However, current multimodal large language models (MLLMs) cannot do this, as they remain largely restricted to static spatial perception, struggling to answer "what-if" questions in a 3D scene. We introduce CausalSpatial, a diagnostic benchmark evaluating whether models can anticipate consequences of object motions across four tasks: Collision, Compatibility, Occlusion, and Trajectory. Results expose a severe gap: humans score 84% while GPT-5 achieves only 54%. Why do MLLMs fail? Our analysis uncovers a fundamental deficiency: models over-rely on textual chain-of-thought reasoning that drifts from visual evidence, producing fluent but spatially ungrounded hallucinations. To address this, we propose the Causal Object World model (COW), a framework that externalizes the simulation process by generating videos of hypothetical dynamics. With explicit visual cues of causality, COW enables models to ground their reasoning in physical reality rather than linguistic priors. We make the dataset and code publicly available here: https://github.com/CausalSpatial/CausalSpatial
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。